Nuclei segmentation and clustering of chondrocytes

Hi everyone,

I have a little issue with a batch of H.E. stained images. The contrast between cell nuclei and matrix isn’t good enough to just set a colour threshold and then run “analyze particle” to get all the nuclei and positions like I normally do. I am interested in getting the position of all the single cells in a cluster of cells without loosing single cells in the normal tissue. Watershed doesn’t really help to separate them and I tried to use the weka segmentation tool (this worked for small images but with my 1 GB files and especially the darker stained sections it didn’t work). Do you have any suggestions what I can try?

thanks in advance,

Dear @Joko,

welcome to the forum! I have fooled around a bit with your images but couldn’t come with a satisfying solution for now. You could take a look at the review of nuclei segmentation methods.

On a more general note, you should upload uncompressed data (if possible, I know that it might be hard with histological images).


Hi Joko,
An alternative would be of course to use a different stain/imaging approach. DAPIi is brilliant for nuclei and an incredibly easy stain to do. Image using a fluorescence microscope and you stand a lot more chance of properly identifying/splitting/counting nuclei. H&E is often not a brilliant stain for image analysis.


1 Like

Thank you Rob,

I decided to do DAPI staining yesterday. I just thought it would be nice to use the H&E images I already had on my computer. But you are right its totally worth the time to use DAPI. I start the first test with DAPI tomorrow before staining the whole batch later this week.

Thanks a lot for your replies.


You could also try Feulgen’s reaction. It is a stoichiometric staining for DNA, used in cell cytometry.